Context Stability
Role-Playing Dimensions: Dates, Regions, and Other Multipliers
Role‑playing dimensions require clear naming and explicit usage to avoid confusion.
TL;DR
- • One dimension can represent multiple roles.
- • AI needs explicit role naming to avoid mix‑ups.
The problem (layman)
- • A single date table is used for order date, ship date, and invoice date.
- • AI does not know which role applies.
Why it matters
- • Role confusion leads to incorrect time comparisons.
- • Answers may mix incompatible contexts.
Symptoms
- • “Last month” results vary by report.
- • Dates referenced in explanations don’t match business intent.
Root causes
- • Role‑playing dimensions not separated or labeled.
- • Measures don’t specify which role is used.
What good looks like
- • Separate role‑playing tables or clear naming for roles.
- • Measures indicate which role they use.
How to fix (steps)
- • Create role‑specific dimensions or views.
- • Rename measures to include date role.
- • Document role usage in metadata.
Pitfalls
- • Assuming users understand role meaning.
- • Mixing roles within one measure.
Checklist
- • Roles clearly defined.
- • Measures indicate role usage.
- • Role‑based tests pass.